Drewett_1999_Field_Archaeology

mean that this is the area of finds research where most time and money should be concentrated. The pottery may be of such a common type that further detailed study may provide little new information. Such material could be studied at a minimal level and archived for future research when perhaps new methods, approaches, or techniques have been developed. Alternatively, samples could be selected for specific analyses that had perhaps not been formerly applied to an otherwise common type.

For practical reasons it is generally better to separate all classes of material into discrete groups for analysis, particularly if specialists in a specific material or artefact type are involved. When the final report is prepared for publication however, the site is what is perhaps more significant than specific types of material. Activity areas involving different materials, or pit groups containing a variety of materials grouped together in a deposit, should be regrouped to show associations which may have had great significance in the past. So although the next section considers discrete types of material, it is their association with other materials, and how and where they were deposited, that may be as important as a detailed analysis of their chemistry or typology.

Finds analysis

Pottery

Once the knowledge of how to make pottery arrives in an area, it is from then on often the dominant find on archaeological excavations. For this reason more detail is given here than for other classes of material. How one deals with the analysis of a pottery assemblage should be based very much on questions to be asked of it and the context of its discovery.An out-of-town rubbish dump will provide a different type of information to potsherds trampled into the mud floor of a house where the pots were used. Sherds from both contexts can, however, be used for dating. Date your potsherds and you at least have a terminus post quem for the date of the deposit in which they were found. Prior to the arrival of Carbon 14 dating, pottery was often used to provide the most reliable chronologies in a region, although as in the case of British prehistory, these chronologies were often too short.

The second major form of information that pottery can provide is about movement of the pots themselves. If you can demonstrate that the clay or materials added to the clay (fillers) used in the pot have a particular geographic source, and the pot is found away from that source, then it is obvious that it must have been taken there, perhaps through trade, gift exchange or movement of the pot s owners. Dating and distribution studies have always dominated archaeological studies of ceramics, but equally important is the function of the pottery in its living context, and what it may indicate about the organization of settlements, and the social, economic, ritual and symbolic life of past societies. Before these considerations are made, the pots, or more likely the sherds from the pots, have to be classified in some way.

The three key elements of pottery analysis are fabric, form and decoration. The form of a pot consists of four main elements: its base, body, neck and rim. If a whole pot is found, all four

148 FIELD ARCHAEOLOGY

Figure 8.1 Thin-sectioning laboratory.

elements can be used to describe its form; but most pottery from archaeological sites is found as sherds, so division by fabric is often considered the most valuable first stage in sorting pottery. It is rare for a single pot to be made of different fabrics, although this could of course happen, particularly with hand-made pottery. The fabric of the pot, sometimes known as its paste, although this term has a more restricted use among potters (Rhodes 1989), consists of the clay matrix and inclusions, tempering or filler. Inclusions , tempering or filler are all terms used for material of little or no plasticity, added to the clay body to control its shrinkage during drying and firing.

The first stage of fabric analysis involves the visual inspection of a fresh break. This may require snipping the sherd with pliers. Some inclusions are readily identifiable by eye or under a binocular microscope, while others may require simple tests, or cutting of thin sections and examinationusingapolarizingmicroscope.Ifinclusions areof areasonablesize,fabrics containing grog (ground down old pots), flint, shell or quartz can be readily identified. Shell, and other calcareousinclusions like limestone,can be confirmed usinga ten per cent solution of hydrochloric acid, while a magnet can confirm the presence of iron compounds. Many rock fragments added to clay by potters cannot be simply identified by eye, so a sample may be needed for analysis.

Petrological analysis requires thin-section samples: to make these, firstly a slice is cut off the sherd using the diamond-edged blade of a stone-cutting saw (Fig. 8.1). The resulting edge is then polished with successively finer particles of corundum: 220 grits, 600 grits and 1,000 grits per cubic millimetre, on flat wet glass surfaces. The corundum particles are then washed off and

Figure 8.2 Ceramic thin section (section and photo: Lys Drewett).

any that have worked their way into the fabric are removed by immersion in an ultrasonic bath. After careful drying, a glass slide is attached to the clean sherd section using hot-cure resin. When the resin is hard, the sherd is mechanically ground to a thickness of approximately 0.01 mm. This section is then hand-polished with corundum particles to a thickness of 0.003 mm, the optimum thickness at which most minerals are rendered translucent or transparent, allowing their optical properties to be studied. There are now also semi-automated systems for grinding and polishing. Finally the new surface is cleaned and sealed with a glass slide cover using resin, which also fills any voids in the specimen to maximise clarity of vision. The slide is then studied under a petrological microscope which has a polarizing light source with a rotating stage.Agood knowledge of optical mineralogy (Kerr 1977) is needed to identify the characteristics of different minerals under polarized light (Fig. 8.2).

Colour and hardness may also be considered when dividing sherds by fabric. Both characteristics are determined more by firing conditions than by the clay or inclusions used by the potter. Colour is best described in relation to a standard colour system rather than using terms to modify basic colours like reddish-brown which, being non-precise, will mean something different to everyone using it. The most commonly used colour system is theAmerican Munsell system which uses colour charts. Each colour is considered to have three variables: hue, intensity and saturation. The hues are the basic colours of the spectrum. Each colour varies in chroma or saturation from black or grey to a clear, pure colour. It also varies in intensity or value from dark tolight.Acolourcanthereforebedescribedby Munsell chart page number andcolour (forexample,

150 FIELD ARCHAEOLOGY

Figure 8.3 Pot elements: base, body, neck and rim.

10YR, that is, page ten, colour yellow-red) followed by a value number (for example, 5) separated from a chroma number (for example, 8) by a slash. A unique colour description to which anyone with a Munsell colour chart can refer would be expressed as, for example 10YR 5/8. A colour may be recorded for both the core of the sherd and its surfaces if they differ.

Hardness also needs a standard if it is to mean anything comparatively. Moh s scale, devised by the German mineralogist Friedrich Moh, is a universal standard. Moh divided hardness of minerals into a scale of ten:

1Talc

2Gypsum

3Calcite

4Fluorite

5Apatite

6Orthoclase

7Quartz

8Topaz

9Sapphire

10Diamond

For pottery hardness, 1 2 on Moh s scale can be scratched by the fingernail so are considered soft. 3 5 cannot be scratched with the fingernail, so are considered hard, while 6 10 are very hard as they cannot be scratched with a steel knife. From the fabric analysis, a type series of fabrics can be established against which new sherds can be compared.

Once sherds have been divided by fabric they can then be divided by form. In sherds this includes four main elements: base, body, neck and rim (Fig. 8.3). Of these generally, but not always, the rim is the most diagnostic, and undecorated body sherds are the least diagnostic. The diagnostic value of necks and bases depends very much on period and geographical origin of the assemblage under study. The bases of early neolithic pottery in Britain for example (they

are all round-based) are not particularly diagnostic, while chronologically significant changes are made to bases in dynastic China.

Rims are one of the most visible elements of a pot in everyday use, so it is hardly surprising that potters varied their forms in every conceivable way. Rims can be made and therefore classified into a wide variety of forms, from plain to thickened, vertical, flaring, incurving, T- shaped, pendant or horizontal. Likewise bases can be divided into round, flat, concave, disk, pointed, and a variety of other regional forms (Joukowsky 1980). Individual body and neck sherds are often difficult to give a form to, but if a whole pot is present it can be given a form description based on geometric solids like spherical, ovoid or cylindrical (Shepard 1956). From these shapes a type-series for the assemblage should be established.

The third key element in pottery analysis is a consideration and classification of decoration. The damp surfaces of newly-made pottery before firing are an ideal medium for decoration by grooving, combing, incising and impressing. Fired surfaces can be painted, while glazes can be used to fire a coating of coloured glass onto the pot s surface. Alternatively, moulded pieces of clay can be added to the surface prior to firing.All these techniques of decoration can be applied in a variety of shapes, some as pure decoration, others providing particular meaning to the people using the pottery. In analysing a decorated pottery assemblage, a type-series of decorative techniques and designs should be established.

The first stage of the analysis of any pottery assemblage therefore involves creating a typeseries of fabrics, forms and decoration. Clearly a single sherd will have a fabric and a form and may have decoration, so the three elements can be combined on a single record form to create a type-series for the whole assemblage (Orton, Tyers and Vince 1993).

The second stage involves attempting to establish how much of which types of pottery are present in a particular assemblage. This involves some form of counting or quantification. The basic problem with this is that archaeologists generally work with sherds, while people in the past used pots. Ideally archaeologists should want to know how many pots of a particular type are present, not how many sherds. However, more often than not, archaeologists have to compare quantities of sherds of different types between areas on a site, between sites, or between periods.

One of the easiest ways of quantifying a pottery assemblage is to count the number of sherds of each type defined in the type series. This will give a rough idea of, say, coarse ware sherds compared with finely decorated sherds, an observation that could probably be made without counting! The problem of counting is that pots of different types break up differently, so sherd counts are biased when attempting to establish proportions of different types. There is also the problem that the degree of fragmentation varies between assemblages (Orton 1975).

An alternative way of quantifying pottery is to weigh sherds of different type. The problem here is that heavy types of pottery, for example flint-gritted storage vessels, will be overrepresented in comparison with lighter types, like grass-tempered wares. Weighing sherds, therefore, also introduces bias, but at least the bias is the same between assemblages. In fact, although problems in quantification by weight are well known, it remains a popular method (Orton, Tyers and Vince 1993).

152 FIELD ARCHAEOLOGY

Given that people in the past used pots rather than sherds, it would clearly be more satisfactory to quantify numbers of pots of different types rather than numbers of sherds. Without whole or fully-reconstructed pots, this can only ever be an estimate. A very rough estimate can be made by grouping together all sherds apparently from the same pot, that is, sherds that appear to be from the same pot but do not actually fit together. This can be used to suggest the minimum number of pots present of a particular type. This type of estimate is so rough that it is of uncertain value.An alternative is to weigh whole pots of each type (if they are available) and divide the weight of sherds of a particular type by the weight of a whole pot of the same type. This again gives a minimum number of pots present.

If the assemblage consists of wheel-turned pottery with round rims, an estimated vesselequivalent (eve) can be calculated (Orton, Tyers and Vince 1993). This involves measuring rim diameters of pots of the same fabric and form using a rim diameter chart. Each rim sherd will be a fraction of a full rim (360°). Therefore by measuring the proportion of the circumference formed by each rim sherd of the same circumference, adding all these together and dividing the total by 360°, you have estimated the minimum number of pots that could be represented (that is, the eve). It remains, however, a minimum number probably many less pots than were actually present in the living context.

Stone

Prior to the introduction of metal, stone was a key tool-making raw material. Its domination in some archaeological assemblages is, however, partly a result of its largely indestructible nature. Tools made of organic material only survive in special circumstances. Stone was also used in the past for a variety of purposes other than making tools. We will start, however, by looking at the analysis of stone tools.

Stone tools can be broadly divided into those that are chipped or flaked and those that are pecked and/or ground. Flaked tools may, however, also be ground or polished, like some neolithic flint axes in Britain. Where available, rocks that fracture concoidally like flint, chert, obsidian and quartz are favoured for flaking, while igneous, some volcanic and some sedimentary rocks are best pecked or ground. The first stage in the analysis of stone tools is to identify the stone used. Broad identification can often be done with a very basic geological knowledge. Flint, obsidian or quartz are readily identifiable, while igneous and sedimentary rocks will require more specialized knowledge. Cutting thin sections, as done for pottery sherds, may be needed to identify rocks petrologically under a polarizing microscope.

Having identified the raw materials being used, the primary division of lithic material is into pieces that have been used, and waste material resulting from making those worked tools. This division requires some basic knowledge of how the tools are made.We will consider chipped and flake tools first. These are generally made of silica like flint, chert, obsidian and quartz, and have a concoidal fracture. Hit these rocks at the right angle and the pressure passes through them as a series of waves, rather like dropping a pebble into water. The flake which breaks off will show these rings or waves radiating out from the point of percussion. Below this point is the bulb of percussion and often a bulbar scar and fissures (Crabtree 1972). Flakes made with a hard

154 FIELD ARCHAEOLOGY

also it can be recycled and, even when buried, most soil conditions will reduce nearly all metals to a corroded mass, making identification difficult or impossible. Only gold, one of the rarest of all metals, survives well in most soils.

The significance of metal finds, and therefore how much effort is worth expending on their analysis, depends very much on the date and context of the find. A bronze age axe perhaps warrants more time than a corroded mass of eighteenth-century nails. The importance of the nails may simply be that they are nails, present in the context in which they were found.

The first stage of analysis is, therefore, to establish what the metal is and what the object was. Visual inspection can usually identify the metal as iron (badly rusted), copper or copper alloy (green corrosion), gold, lead or silver.Alloys are more of a problem, and require laboratory analysis to identify trace elements. Optical emission spectrometry, atomic absorption spectrometry, or x-ray fluorescence can all be used to establish the composition of an alloy.

If not badly corroded, identification of the object may be straightforward: an axe, a knife or a coin. If corroded, the object may be unidentifiable by surface observation. Often the object survives in a casing of corrosion, so can be revealed by x-ray, which is far quicker and cheaper than full conservation. Fragments of objects are, however, more difficult to identify except by a specialist in the field. If metal appears in any quantity, types should be defined and numbers counted by context.

Organic artefacts

In most periods in the past, many artefacts were made of organic materials like wood, skin, bark, feathers and bone. The preservation of organic materials requires specific burial conditions, so most archaeological sites reveal no organic materials unless the objects are carbonized. Except on waterlogged or desiccated sites, one must presume a strong bias against organic finds and in favour of inorganic materials like stone and pottery.

Wooden objects will survive only if waterlogged, desiccated or carbonized, although an indication of wood may also survive when buried against metal objects, as the corrosion products can replace the wood. Most organic materials require conservation before analysis can begin. Until recently the most common form of wood conservation was to replace the water content of the object with a solution of polyethylene glycol (PEG), but now freeze-drying is often used for smaller artefacts (Dowman 1970).

Having stabilized the wooden object, the first stage of analysis is to identify the wood. This may involve cutting a cube out of the object to examine transverse, radial longitudinal and tangential longitudinal sections (Dimbleby 1978). This enables the cellular structure of the wood to be examined under a lens or microscope (Fig. 8.4). Each species of wood has its own structure (Jane 1956). The next stage of analysis is a consideration of how the object was made, by identifying traces of cutting, scraping, carving, polishing and in later periods planing. Finally the nature and possible use of the wooden object should be determined if possible. Some objects like bows, arrow shafts, pins and axe handles are reasonably easy to identify while others, especially in fragmentary form, require specialized knowledge (Taylor 1981).

Figure 8.4 Cellular structure of wood under a microscope.

Bone artefacts survive in a wider range of soil conditions than wooden ones. Bone consists of a combination of calcium and collagen. The collagen element survives well in waterlogged acidic conditions, while the calcium element survives in many dry calcareous soils. As with wood, the post-excavation analysis should proceed from conservation (if required) to species identification, manufacturing process, and functional identification. Other organic materials like basketry, textiles, skin, bark and feather artefacts are rare finds and generally require specialist conservation and identification.

Artefacts of other materials like glass, amber and ivory all require specialist conservation treatment prior to analysis by finds specialists in these fields.

Finds analysis: ecofacts

The analysis of ecofacts and environmental samples at the post-excavation stage is usually undertaken by specialists in these fields rather than by the field archaeologist who directed the excavation, although not infrequently the latter may have expertise in these areas. The field archaeologist s role may therefore be restricted to safely packaging and delivering samples to the specialist, providing detailed contextual information, and incorporating the results of the analyses into the final archive and published report.

The term ecofact is generally taken to cover all non-artefactual materials which have been modified by humans, and so have cultural relevance. This would cover food refuse like animal bones and shells from shellfish, but also pollen, land snails and humanly-modified soils. I propose in this section, however, to separate evidence that provides economic data from that

156 FIELD ARCHAEOLOGY

which essentially provides environmental data, although clearly the two overlap and are interrelated.

Bones Bones usually survive well in soils that are not too acidic, but they must have been buried fairly quickly to avoid attrition by scavengers and natural erosion through weathering. The collagen element of bone will, however, survive well in acidic waterlogged conditions, as in the case of the bog people of Europe. Food refuse bones from excavations will probably have been recovered in slightly different ways. Bird and fish bones require fine mesh sieving (1 mm) while large mammal bones are generally recovered through trowelling, perhaps with coarse mesh sieving (1 cm). Bones should be carefully cleaned with water and a soft brush prior to analysis. Fragile bones may require dry brushing or consolidation with a dilute solution of polyvinyl acetate (PVA).

The first stage of bone analysis is to identify what species, and what part of the body, each bone comes from. This is best done by use of a comparative reference collection, although a start can often be made by consulting illustrations (Ryder 1968). Fragmented bones, particularly shaft splinters, are naturally more difficult to identify than whole bones. A proportion of any bone assemblage may not be identifiable to species level.

Having identified the bones to species and to part of the skeleton, the bone assemblage should be quantified. As with pots and potsherds, the archaeologist should be more interested in how many whole animals are represented than how many bones are represented. A simple count of the number of identified species can give a very biased picture, although this may be appropriate if very small numbers are present. Generally a count of minimum number of individuals (MNI) is undertaken to compare abundance of species between contexts, periods and sites. The minimum number of individuals is calculated by counting only those bones which appear once in an animal, like a left scapula. If 100 left ox scapulae are present, the minimum number of oxen present is 100. This can then be translated into protein available to the people herding or hunting the various species by calculating meat weight represented by each animal. Zooarchaeology is now a significant subdiscipline of archaeology, so more complex analyses are generally beyond the expertise of the field archaeologist (Chaplin 1971). Sites crucial for the understanding of animal domestication, for example, would almost certainly have involved a zooarchaeologist in the initial planning, as well as the field element and post-excavation stages of a project.

Shells Thepreliminary analysis of shells from archaeological excavations can beapproached in a similar way to the analysis of bones, although the excavation of shell middens requires rigorous sampling in the fieldtokeepthe jobto manageable proportions. Shellssurviveremarkably well and, if middened, can even survive on quite acid soils which would otherwise dissolve calcareous material.

Sites of many periods, particularly near coastlines or rivers, produce shells mixed with other domestic rubbish. Care should be taken to attempt to sort out shells which could be food